how to backtest model estimates

For econometric discussions not necessarily related to EViews.

Moderators: EViews Gareth, EViews Moderator

sidddigobaid
Posts: 4
Joined: Tue Jun 04, 2013 5:58 am

how to backtest model estimates

Postby sidddigobaid » Fri Jun 07, 2013 6:37 am

As you may know in order to evaluate the quality of estimates, the model estimate/results should always be backtested with appropriate methods!! the question here is what methods should I apply to backtest my etimates/coefficients?
Thank you in advance for your help

trubador
Did you use forum search?
Posts: 1520
Joined: Thu Nov 20, 2008 12:04 pm

Re: how to backtest model estimates

Postby trubador » Fri Jun 07, 2013 3:03 pm

Actually, backtesting is used to test the validity of a strategy to be implemented, which is quite a common concept among traders. It also (sometimes) refers to testing the predictive ability of the model at hand using the past data. However, validity of estimates or coefficients do not necessarily refer to the same thing in econometrics so there are various diagnostic tools to serve the purpose. For instance, EViews offers many of such tools (for univariate models) under three different categories: 1) Coefficient diagnostics, 2) Residual diagnostics and 3) Stability diagnostics.

sidddigobaid
Posts: 4
Joined: Tue Jun 04, 2013 5:58 am

Re: how to backtest model estimates

Postby sidddigobaid » Sat Jun 08, 2013 4:12 am

That's fine. I'm currently using VAR model to assess the impact of oil prices fluctuations on some macroeconomic indicators. I got number of coefficients which show me that oil price fluctuations have influence on GDP growth and inflation (significant). In this case how to validate this results? what tool should I use to cross-check the validity of this outcomes.


Return to “Econometric Discussions”

Who is online

Users browsing this forum: No registered users and 2 guests